Model reference · open weights
NVIDIA-Nemotron-3.5-Transcribe is an open-weight audio or speech model from NVIDIA. NVIDIA-Nemotron-3.5-Transcribe-30B-A4B (BF16) weighs 67.5 GB; the smallest configuration that runs it is H100 80 GB.
For self-hosting, compare this build's requirements below with the GPU you plan to rent.
Summary of the nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B model card, 2026-10-30
What it is
| Released by | NVIDIA |
|---|---|
| Released | 2026-10-07 |
| Parameters | 33.7B |
| VRAM | 67.5 GB for the weights |
What it runs on
| Card | One stream | Memory |
|---|---|---|
| RTX 3060 12 GB … L40S 48 GB | does not fit | |
| A100 80 GB | tight | 78.2 GB |
| H100 80 GB | tight | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | fits | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | fits | 107 GB |
| H200 141 GB | fits | 138 GB |
| B200 180 GB | fits | 176 GB |
From the model card
Nemotron 3.5 Transcribe is an open-weight, offline speech and audio model for transcription in 40 languages, with support for recordings up to one hour and a 65,536-token context. Its hybrid mixture-of-experts architecture activates approximately 4.3B parameters, including the audio encoder, and it is released in BF16, FP8, and NVFP4 precision.
Get started ↓ · Explore the tasks ↓ · See the evidence ↓ · NeMo Speech ↗
NVIDIA Nemotron-3.5-Transcribe is an offline speech and audio model that processes recordings, together with text instructions and optional context, into text responses.
NVIDIA developed this Speech-Augmented Language Model (SALM) for multilingual transcription, speech translation, contextual transcription, speaker-attributed transcription, rich transcription, timestamp prediction, language identification, and audio question answering. The training mixture also includes a small amount of text instruction following and PII tagging/redaction tasks.
This model is ready for commercial or non-commercial use.
| Total Parameters | 34B (4.3B active), including the audio encoder |
| Architecture | Mamba-2–Transformer hybrid Mixture of Experts language model with a Parallel Expert Encoder (Transformer ASR encoder and speaker encoder) |
| Precision | BF16 (full-precision reference weights); FP8 and NVFP4 variants are available |
| Context Length | 65,536 tokens; recordings up to 1 hour |
| Inputs / Outputs | Audio and text / text, with optional word timestamps |
| Languages | 40; see the language support matrix |
| Tasks | Transcription in verbatim or written form, contextual transcription, speaker-attributed transcription, rich transcription, speech translation, language identification, and audio question answering |
| Deployment | 1× H100 80 GB; see the hardware compatibility table |
| Speculative Decoding | Native Multi-Token Prediction (MTP); optional DFlash2 draft model |
| Reasoning Mode | Configurable through the chat template (enable_thinking=True/False) |
| Recommended Sampling | Transcription, translation, and language identification: temperature 0, reasoning off; audio understanding: temperature 0.6, top_p 0.95, reasoning on |
| Best For | Highest accuracy, customization, and fine-tuning |
| License | OpenMDW-1.1 |
Results captured on October 12, 2026. Error rates are percentages. ↓ Lower is better. Complete results are listed under Benchmark Results.
Best published results
| Capability | Benchmark | Metric | Nemotron 3.5 Transcribe | Next best published |
|---|---|---|---|---|
| Multi-speaker ASR | AMI, CALLHOME, Mixer 6, NOTSOFAR-1 full sessions | Mean cpWER ↓ | 15.80 | DiCoW v3.3 + DiariZen 16.83 |
| Style control | Preference-ASR | WER ↓, default / instructed | 7.56 / 6.16 | Canary-Qwen 8.16 / 8.36 |
Best open-weight results
| Capability | Benchmark | Metric | Nemotron 3.5 Transcribe | Next best open-weight model |
|---|---|---|---|---|
| English ASR | Open ASR Leaderboard, 9 public sets | Mean WER ↓ | 4.71 | Qwen3-ASR-1.7B 4.94 |
| Long-form ASR | Open ASR Leaderboard, long-form | Mean WER ↓ | 10.98 | Parakeet-TDT-0.6B-v3 13.37 |
| Multilingual ASR | Multilingual Open ASR Leaderboard, 7 languages | Mean WER ↓ | 4.62 | Qwen3-ASR-1.7B 6.49 |
With the optimized vLLM implementation in the Nemotron Transcribe NIM, one RTX PRO 6000 Blackwell Server Edition GPU transcribes up to 2,238 seconds of audio per second (NVFP4 with DFlash2 speculative decoding) and returns the transcript of a 30-second recording in 0.13 seconds. These serving optimizations are not yet part of the open-source release; we aim to upstream them to NeMo Speech and vLLM at a later date. The vLLM integration shipped in NeMo Speech on the release day is intended for model development and is expected to be roughly two times slower.
The example below serves this checkpoint on one H100 80 GB with vLLM 0.32.0 and the latest NeMo Speech main branch. Install both and download the checkpoint as described in Environment Setup.
export VLLM_PLUGINS=nemo_speechlm
export VLLM_MAX_AUDIO_DECODE_DURATION_S=3600
export VLLM_MAX_AUDIO_CLIP_FILESIZE_MB=180
# Word timestamps require the V1 model runner.
export VLLM_USE_V2_MODEL_RUNNER=0
vllm serve ./NVIDIA-Nemotron-3.5-Transcribe-30B-A4B \
--served-model-name nvidia/NVIDIA-Nemotron-3.5-Transcribe-30B-A4B \
--trust-remote-code --dtype bfloat16 --gpu-memory-utilization 0.94 \
--max-model-len 65536 --max-num-batched-tokens 4096 --max-num-seqs 192 \
--enable-chunked-prefill --no-enable-prefix-caching \
--use-replayssm --mamba-backend flashinfer --mamba-cache-mode none \
--mm-proQuoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.